docs: dpmodel, model conversion (#3360)

Signed-off-by: Jinzhe Zeng <[email protected]>

doc/backend.md

@@ -9,12 +9,33 @@ In the documentation, TensorFlow {{ tensorflow_icon }} and PyTorch {{ pytorch_ic
 
 ### TensorFlow {{ tensorflow_icon }}
 
-TensorFlow 2.2 or above is required.
+- Model filename extension: `.pb`
+- Checkpoint filename extension: `.meta`, `.index`, `.data-00000-of-00001`
+
+[TensorFlow](https://tensorflow.org) 2.2 or above is required.
 DeePMD-kit does not use the TensorFlow v2 API but uses the TensorFlow v1 API (`tf.compat.v1`) in the graph mode.
 
 ### PyTorch {{ pytorch_icon }}
 
-PyTorch 2.0 or above is required.
+- Model filename extension: `.pth`
+- Checkpoint filename extension: `.pt`
+
+[PyTorch](https://pytorch.org/) 2.0 or above is required.
+While `.pth` and `.pt` are the same in the PyTorch package, they have different meanings in the DeePMD-kit to distinguish the model and the checkpoint.
+
+### DPModel {{ dpmodel_icon }}
+
+:::{note}
+This backend is only for development and should not take into production.
+:::
+
+- Model filename extension: `.dp`
+
+DPModel is a reference backend for development, which uses pure [NumPy](https://numpy.org/) to implement models without using any heavy deep-learning frameworks.
+Due to the limitation of NumPy, it doesn't support gradient calculation and thus cannot be used for training.
+As a reference backend, it is not aimed at the best performance, but only the correct results.
+The DPModel backend uses [HDF5](https://docs.h5py.org/) to store model serialization data, which is backend-independent.
+Only Python inference interface can load this format.
 
 ## Switch the backend
 
@@ -26,3 +47,7 @@ When training and freezing a model, you can use `dp --tf` or `dp --pt` in the co
 
 When doing inference, DeePMD-kit detects the backend from the model filename.
 For example, when the model filename ends with `.pb` (the ProtoBuf file), DeePMD-kit will consider it using the TensorFlow backend.
+
+## Convert model files between backends
+
+If a model is supported by two backends, one can use [`dp convert-backend`](./cli.rst) to convert the model file between these two backends.

doc/conf.py

@@ -188,6 +188,7 @@ def setup(app):
 myst_substitutions = {
     "tensorflow_icon": """![TensorFlow](/_static/tensorflow.svg){class=platform-icon}""",
     "pytorch_icon": """![PyTorch](/_static/pytorch.svg){class=platform-icon}""",
+    "dpmodel_icon": """![DPModel](/_static/logo_icon.svg){class=platform-icon}""",
 }
 
 # -- Options for HTML output -------------------------------------------------

doc/model/train-se-e2-a.md

@@ -1,7 +1,7 @@
-# Descriptor `"se_e2_a"` {{ tensorflow_icon }} {{ pytorch_icon }}
+# Descriptor `"se_e2_a"` {{ tensorflow_icon }} {{ pytorch_icon }} {{ dpmodel_icon }}
 
 :::{note}
-**Supported backends**: TensorFlow {{ tensorflow_icon }}, PyTorch {{ pytorch_icon }}
+**Supported backends**: TensorFlow {{ tensorflow_icon }}, PyTorch {{ pytorch_icon }}, DPModel {{ dpmodel_icon }}
 :::
 
 The notation of `se_e2_a` is short for the Deep Potential Smooth Edition (DeepPot-SE) constructed from all information (both angular and radial) of atomic configurations. The `e2` stands for the embedding with two-atoms information. This descriptor was described in detail in [the DeepPot-SE paper](https://arxiv.org/abs/1805.09003).

doc/model/train-se-e2-r.md

@@ -1,7 +1,7 @@
-# Descriptor `"se_e2_r"` {{ tensorflow_icon }}
+# Descriptor `"se_e2_r"` {{ tensorflow_icon }} {{ pytorch_icon }} {{ dpmodel_icon }}
 
 :::{note}
-**Supported backends**: TensorFlow {{ tensorflow_icon }}
+**Supported backends**: TensorFlow {{ tensorflow_icon }}, PyTorch {{ pytorch_icon }}, DPModel {{ dpmodel_icon }}
 :::
 
 The notation of `se_e2_r` is short for the Deep Potential Smooth Edition (DeepPot-SE) constructed from the radial information of atomic configurations. The `e2` stands for the embedding with two-atom information.